The broad goals of this project are to understand the scope and sophistication of retinal processing. This project builds on preliminary data showing that the retina can quickly learn simple patterns in the visual stimulus, such as a temporal periodicity or a smooth motion trajectory, so that a subset of ganglion cells fires selectively after a violation of the predicted pattern. Such cognitively loaded phenomena are typically thought to arise only in the cortex, but observing them in the retina both enhances our appreciation of the computational capabilities of neural circuits and also allows us to explore the mechanisms responsible, a task that often cannot be carried out in more central neural circuits. The Specific Aims are: 1) exploring the properties of retinal pattern detection with multi-electrode array experiments;2) studying the mechanisms of temporal pattern detection using designed stimuli, pharmacology, and computational models;3) studying the properties of and mechanisms behind the retina's ability to detect motion discontinuities. A detailed knowledge of how the population of retinal ganglion cells represents the visual world is of fundamental interest to neuroscience and is also important for guiding the development of a retinal prosthesis to restore vision successfully.